ESX 3.5 Standard - HA

[fzilz]

I have ESX 3.5 Standard Acceleration pack. I have 2 hosts in 1 cluster. I have HA enabled and my Virtual Machines exist on a datastore that is an iSCSI LUN (on a NS 350 Celerra). Both Hosts see the data store and when the VMs are powered off, I can quickly migrate from one host to the other and come backup. My Infrastructure Virtual Center Server is on a VM.

I am having trouble getting a clear picture of when and how HA will automatically migrate my VM's in the event of a host failure or loss of connectivity. and If the failure occurs on teh host which is running the Virtual Center Server how that will work. Can someone point me to some documentation for how HA works for the Standard license not enterprise.

[apatel1]

As you said, HA will kick in when there is lost of connectivity or if there is a host failure. If there is loss of connectivity, the hosts use an "isolation address" to determine which host is actually the "bad" one. In a cluster of three hosts, for example, if one host suddenly loses connectivity, it doesn't necessarily know if it has lost connectivity, or if all of the other hosts did. In this case, they all verify connectivity by attempting to reach the "isolation address" which, by default, is the Service Console default gateway address. If a host determines that it is the one that is isolated, it will either leave the VMs powered on or forcefully power them off (depending on what you have the Isolation Response set to) and the other hosts will attempt to power them on.

In your case, where VirtualCenter is a VM, that is not an issue, as HA does not depend on VirtualCenter to do its work. If you lose the host running the VirtualCenter VM, HA will still function as it normally would.

Hope that helps! Please help me out by marking my response as "helpful" or "correct" if you feel that it was useful!

-Amit

[joms_marino]

Hi Amit,

I'm just confuse with what you say, but I think I have read some article that you need VC for the HA to work. You states here that even if the VC running on a VM which resides on the ESX Server that gets down HA will stil work. *"In your case, where VirtualCenter is a VM, that is not an issue, as HA

does not depend on VirtualCenter to do its work. If you lose the host

running the VirtualCenter VM, HA will still function as it normally

would."*

Please clarify since I'm just new also with this technology.

Thanks and Regards,

Joms

[madda]

VC is used to setup HA, but once it is setup it runs between the ESX hosts rather than using VC. So if VC goes down, HA will still work once it is configured

[TomHowarth]

Firstly HA works exactly the same way in Standard as it does in Enterprise there any documentation that relates to HA is the same.

Secondly HA only required vCenter for the creation of the HA Cluster. once it is active the Hosts themselves manage it. the heartbeat is monitored at a Host level and if any host fails one host in the cluster will start the guests, however, without DRS, you will need to manage that procedure as you could end up with a cascading failure, especially if you are aggressive with your Guest levels. you need to make sure that your HA cluster can handle a Host failure and still be able to restart your guests. Remember .N=1

If you found this or any other answer useful please consider the use of the Helpful or correct buttons to award points

Tom Howarth

VMware Communities User Moderator

Blog: www.planetvm.net

[rminick]

I am searching for something and thought I'd chime into this thread with it. Is the loss of connectivity based on ICMP/Ping to the gateway? The reason I ask is we had a ESX server fail today to some degree but HA didn't kick in. There was a storage issue locally on the host which had it locked up on IO Errors galore. It was still pingable so could that be why HA didn't bring up the 2 isolated VM's on remaining hosts? Just want to make sure there's not an advanced option to do a bit more checking if this is the case.

Richard